lip-sync-spokesperson-video
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to ingest and process untrusted external media, such as face images and audio files, which creates a surface for indirect prompt injection attacks.
- Ingestion points: The
generate_lip_synctool acceptsface_imageandaudioparameters from user-provided data (SKILL.md). - Boundary markers: There are no explicit boundary markers or instructions to ignore potential commands embedded in the provided media metadata or content.
- Capability inventory: The skill uses the
generate_lip_syncandget_video_generationtools to perform remote video generation tasks. - Sanitization: No sanitization, validation, or filtering of the input media files is mentioned to prevent the processing of adversarial instructions.
- [COMMAND_EXECUTION]: The skill documentation includes a command for setting up the Model Context Protocol (MCP) transport layer, which involves executing a shell command.
- Evidence:
claude mcp add --transport http apimage https://mcp.apimage.org/mcp --header "Authorization: Bearer sk_your_api_key" -s userin SKILL.md. - Note: This command connects to a service hosted on
apimage.org, which is the domain associated with the skill vendorapimageorg.
Audit Metadata